Open Access
Add to BibSonomy
Abstract
The particle composition of suspended matter provides crucial information for a deeper understanding of marine biogeochemical processes and environmental changes. Particulate backscattering efficiency (Qbbe(λ)) is critical to understand particle composition, and a Qbbe(λ)-based model for classifying particle types was proposed. In this study, we evaluated the applicability of the Qbbe(λ)-based model to satellite observations in the shallow marginal Bohai and Yellow Seas. Spatiotemporal variations of the particle types and their potential driving factors were studied. The results showed that the Qbbe(λ) products generated from Moderate Resolution Imaging Spectroradiometer (MODIS) on the satellite Aqua agreed well with the in situ measured values, with determination coefficient, root mean square error, bias, and mean absolute percentage error of 0.76, 0.007, 16.5%, and 31.0%, respectively. This result verifies the satellite applicability of the Qbbe(λ)-based model. Based on long-term MODIS data, we observed evident spatiotemporal variations of the Qbbe(λ), from which distinct particle types were identified. Coastal waters were often dominated by minerals, with high Qbbe(λ) values, though their temporal changes were also observed. In contrast, waters in the offshore regions showed clear changes in particle types, which shifted from organic-dominated with low Qbbe(λ) levels in summer to mineral-dominated with high Qbbe(λ) values in winter. We also observed long-term increasing and decreasing trends in Qbbe(λ) in some regions, indicating a relative increase in the proportions of mineral and organic particles in the past decades, respectively. These spatiotemporal variations of Qbbe(λ) and particle types were probably attributed to sediment re-suspension related to water mixing driven by wind and tidal forcing, and to sediment load associated with river discharge. Overall, the findings of this study may provide valuable proxies for better studying marine biogeochemical processes, material exchanges, and sediment flux.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Suspended particulate matter (SPM), such as phytoplankton, detritus, and minerals, is an important component of oceans and seas because of its significant role in the determination of marine biogeochemical processes [1,2]. For instance, SPMs are considered essential carriers of nutrients and marine pollutants [3,4]. Sunlight is inevitably absorbed or scattered by SPM when it penetrates the water column, affecting phytoplankton productivity, and thereby marine ecosystem [5,6]. Owing to the hydrographic dynamics of oceanic and coastal waters, SPM often undergoes significant changes [7,8]. Thus, it is also regarded as a measure of water clarity and even an indicator of environmental patterns related to climate change [9,10].
SPM usually has a complex and variable composition, and can be generally classified into two types: particulate organic matter (POM) and particulate inorganic matter (PIM) [11]. POM commonly includes phytoplankton and their detritus, whereas PIM generally includes inorganic sediments [12,13]. Different types of SPM have distinct functions in the marine biogeochemical processes. As POM is mainly generated by phytoplankton photosynthesis, it is a vital carbon pool and forms the base for studying marine carbon cycles [14,15]. Furthermore, as a food source for aquatic organisms, POM provides essential information on evaluating the relative amount of heterotrophy [16,17]. PIM is commonly introduced by river discharges or re-suspension of lithogenic sediments from the seafloor, and can be used to study material transport, sediment flux, etc. [18,19]. Therefore, understanding the composition of SPM is essential for a deeper understanding of the marine environment and biogeochemical processes.
Several traditional laboratory analysis methods exist for determining the SPM composition, such as the thermogravimetry, pressure-calcimeter, and weight loss-on-ignition methods [20–22]. Although these methods are precise, they are often time- and labor-intensive. More importantly, the information provided by them is spatially limited by the number of discrete and sparse observation stations, which hinder the understanding of the spatiotemporal variations of SPM composition. Compared with traditional laboratory methods, satellite remote sensing observations have broad spatial coverage and frequent intervals, and are widely used to study water environment changes.
As the optical basis of remote sensing model, the inherent optical properties, especially the scattering characteristics of SPM related to particle composition, have been extensively examined [23–28]. Researchers reported that the particulate backscattering ratio (Bbp(λ)), which is defined as the ratio of scattering (bp(λ)) to backscattering coefficients (bbp(λ)) of particles, is strongly dependent on particle composition. For instance, based on Mie theory, Twardowski et al. [23] proposed a model to estimate the bulk refractive index from Bp(λ), and classified particles in the Gulf of California into three groups according to its levels. In their review of optical techniques for remote and in situ characterization of particles in the GEOTRACES program, Boss et al. [26] recommended that Bp(λ) is a useful descriptor of particle composition. However, it must be noted that the Bp(λ)-based methods may only be applied to in situ data, but not to satellite observations as it is theoretically infeasible to derive bp(λ) from the remote sensing reflectance (Rrs(λ)) [29,30].
In addition to Bp(λ), Wang et al. [31] reported that backscattering efficiency (Qbbe(λ)), which is the ratio of bbp(λ) to the cross-sectional area concentration of particles (CSA), could effectively derive particle types in Chinese coastal waters. The Qbbe(λ) levels varied >30-fold and clustered into two types, which are dominated by large organic and small mineral particles, respectively. Furthermore, Wang et al. [32] reported that the division of Qbbe(λ) is seasonally stable and proposed a Qbbe(λ)-based method for particle classification. It should be noted that satellite bbp(λ) products are available, which can be produced using an inherent optical algorithm [29,30]. Meanwhile, CSA has also been successfully derived from satellite Rrs(λ) using the empirical remote sensing algorithm proposed by Wang et al. [33]. These studies indicate that the Qbbe(λ)-based method for deriving SPM composition has great potential for application to satellite observations. However, up to now, satellite estimation of particle composition using the Qbbe(λ)-based method is still lacking, which limits our understanding of the spatiotemporal variations in particle composition.
To fill the above research gaps, this study aims to apply the Qbbe(λ)-based model to satellite observations of Moderate Resolution Imaging Spectroradiometer (MODIS) on the satellite Aqua in two typical shallow marginal seas: the Bohai Sea (BS) and Yellow Sea (YS). The main objectives were summarized as: (1) evaluate the satellite applicability of Qbbe(λ)-based model based on in situ measurements; (2) use long-term MODIS data to reveal the spatiotemporal variation patterns of Qbbe(λ) and the particle types; (3) discuss potential environmental driving factors that are responsible for these variations.
2. Materials and methods
2.1 Study area and sampling
The BS, with an area of ∼77,000 km2 and an average depth of 18 m, is a typical semi-enclosed sea [34]. The YS, located between mainland China and the Korean Peninsula and connected to the BS through the Bohai Strait, is larger and deeper than the BS, with an area of ∼380,000 km2 and an average depth of ∼70 m [34]. These two relatively shallow seas are heavily affected by freshwater in summer and strongly influenced by monsoons in winter [35]. The BS receives a large amount sediment from the Yellow River, the second largest sediment-loaded river in the world with an annual runoff of 890 × 108 m3 [36]. Owing to the monsoons, river discharges, and tidal forcing, wide regionally and seasonally varying water properties have been observed in the BS and YS [27,37].
This study first validated the satellite-derived bbp(λ), Qbbe(λ), and CSA based on in situ measurements. These data were collected from three cruises conducted in the BS and YS during winter (December 2016), spring (April 2018), and summer (July 2018) aboard the R/V Dongfanghong2 (Fig. 1). A profiling package, including Sequoia Scientific LISST-100X (type C), HOBI Labs Hydroscat-6, and Seabird SBE911P conductivity-temperature-depth profiler, was used to measure the optical and hydrological properties of water, i.e., particle size distribution, light backscattering, and water column temperature and salinity [32].
Fig. 1. Study region and sampling locations of the match-ups of in situ measurements and satellite observations in the Bohai Sea (BS) and the Yellow Sea (YS) during December 2016, April 2018, and July 2018. Areas with different green backgrounds indicate the BS, north YS, and south YS. Red rectangles denote eight selected sub-regions.
To investigate the spatiotemporal variations of particle backscattering efficiency and types in a specific region, the study region was divided into three parts: the BS, North Yellow Sea (NYS), and South Yellow Sea (SYS) (Fig. 1). For the same purpose, we also selected eight typical sub-regions: the central BS (R1), Bohai Strait (R2), Yellow River Estuary (R3), central NYS (R4), central SYS (R5), Subei Shoal (R6), Yangtze Estuary (R7), and East China Sea (ECS) shelf (R8).
2.2 In situ data measurements
Total backscattering coefficients (bb(λ)) at 442, 488, 550, 640, 700, and 852 nm were determined using the Hydroscat-6 of HOBI Labs, which measures the total volume scattering function at a fixed angle of ∼140° in the backward direction [38]. The particulate backscattering coefficient bbp(λ) was obtained by subtracting the backscattering coefficient of pure water from bb(λ). The CSA values were derived using the LISST-100X Type-C particle size analyzer (Sequoia Scientific Inc.), which measures light scattering produced by suspended particles in a volume of water in 32 logarithmically spaced scattering angles in a near-forward direction [39]. Detailed calculations of CSA values from LISST measurements can be found in Wang et al. [32]. Consequently, the backscattering efficiency of the particles, Qbbe(λ), was determined as
2.3 Satellite data
To validate the satellite-derived bbp(λ), CSA, and Qbbe(λ), we collected MODIS-Aqua L2 daily Rrs(λ) products during the three cruises, December 2016, April 2018, and July 2018, from the NASA Ocean Color website (https://oceancolor.gsfc.nasa.gov/). The Rrs(λ) data were derived using the standard atmospheric correction by Ocean Biology Processing Group (OBPG) at NASA. These data were then used to build a match-up dataset with synchronous in situ measurements. In total, 41 data points were strictly matched between the MODIS daily Rrs(λ) L2 products and in situ measured bbp(λ), CSA, and Qbbe(λ) with a spatial tolerance of 3 × 3 pixel and an overpass temporal window of ±6 h. To reduce the influence of outliers, we used the median Rrs(λ) values from the 3 × 3 pixel window centered on the locations of the field sampling stations as the MODIS satellite Rrs(λ). Furthermore, MODIS satellite Rrs(λ) was used to derive bbp(λ) using the quasi-analytical algorithm (QAA, version 6) [29,40], and CSA values were calculated using the empirical algorithm of Wang et al. [33]. Finally, the satellite Qbbe(λ) values were calculated using Eq. (1).
We also obtained MODIS-Aqua L3 monthly Rrs(λ) products from the NASA Ocean Color website (https://modis.gsfc.nasa.gov/data/), which were used to explore the spatiotemporal variations of Qbbe(λ) and particle composition in the BS and YS. Monthly bbp(λ), CSA, and thereby Qbbe(λ) were calculated using the methods described above. To investigate the potential driving factors of Qbbe(λ) and particle composition, we also computed MODIS satellite products of the chlorophyll a (Chla) concentration and total suspended matter (TSM) using the Yellow Sea Ocean Color (YOC) algorithm [41], which was developed specifically for the YS by the Yellow Sea Large Marine Ecosystem Ocean Color Work Group. The data and methodology used in this study were shown in Fig. 2.
2.4 Qbbe(λ)-based method for determining particle types
Based on in situ data collected in the BS and YS during several cruises in different seasons, Wang et al. [31,32] observed large variations in the Qbbe (Table 1) and reported that the frequency distribution of the Qbbe at 640 nm had two clear clusters with low and high mean values. Each cluster was approximately log-normally distributed and dominated by POM and PIM, respectively. Based on the frequency distribution of the Qbbe(640) data, Wang et al. [32] determined a threshold of the Qbbe(640) data with a value of 0.0134 to classify suspended particles into two types: Type 1 particles with low Qbbe(640) levels are dominated by POM, type 2 particles with high Qbbe(640) values have high proportions of PIM. Most current ocean color sensors do not have a 640 nm band, whereas 550 nm is a ubiquitous band. Analysis of the relationship between Qbbe(640) and Qbbe(550) showed a strong correlation between them, with a determination coefficient (R2) value of 0.99 (Fig. 3). Based on this relationship (fitted function), a threshold of 0.0157 for Qbbe(550) was adjusted to classify the particle types from satellite measurements. To apply the Qbbe(λ)-based method to MODIS satellite data, we used the Qbbe at 547 nm band of MODIS sensor as the replacement band of 550 nm to classify suspended particles, i.e., when Qbbe(547) > 0.0157, the particles were dominated by PIM, and when Qbbe(547) ≤ 0.0157, the particles were dominated by POM.
Fig. 3. Relationship between Qbbe(640) and Qbbe(550). Solid blue line indicates the fitted function.
Table 1. Statistic of the Qbbe at 550 and 640 nm in the Bohai Sea and Yellow Sea in terms of maximum (Max), minimum (Min), mean, standard deviation (SD), and coefficient of variation (CV).
2.5 Accuracy assessment
We used four accuracy indicators, R2, root mean square error (RMSE), mean of percent error (bias), and mean of absolute percent error (MAPE) to evaluate the quality of bbp(λ), CSA, and Qbbe(λ) derived from MODIS satellite observations based on the match-up dataset with synchronous in situ measurements. The RMSE, bias and MAPE were calculated as follows [42]:
3. Results
3.1 Validation of satellite-derived backscattering efficiency of particles
Based on the match-up dataset (N = 41) of in situ measurements and MODIS observations, we assessed the accuracy of MODIS-derived bbp(547), CSA, and Qbbe(547). The bbp(547) values calculated from MODIS Rrs(λ) using QAA agreed well with the in situ measured values, with R2, RMSE, bias, and MAPE values of 0.86, 0.047 m−1, 0.2%, and 18.7%, respectively (Fig. 4). A strong relationship between MODIS satellite-derived CSA and in situ measured values was also observed, showing that most samples clustered around the 1:1 line. The R2, RMSE, bias, and MAPE values were 0.78, 2.49 m−1, −5.3%, and 30.7%, respectively. Consequently, the satellite-derived Qbbe(547) showed good consistency with the in situ measured data, with R2, RMSE, bias, and MAPE values of 0.76, 0.007, 16.5%, and 31.0%, respectively. These results indicate that the satellite products of Qbbe(547) generated from MODIS observations are of good quality and can be applied to study the spatiotemporal variation patterns of Qbbe(547), and thereby the particle types in the BS and YS.
Fig. 4. Comparisons between in situ measured and MODIS derived bbp(547) (a), CSA (b), and Qbbe(547) (c). Solid red lines indicate the 1:1 line. Solid blue lines denote the type II linear regression line between log-transformed in situ data and MODIS-derived values.
3.2 Spatial variations of backscattering efficiency and particle types
Using the MODIS L3 monthly Rrs(λ) products, the average monthly products of Qbbe(547) in the BS and YS during 2003–2018 were obtained (Fig. 5). According to the Qbbe(λ)-based method of particle classification, type 1 particles with low Qbbe(547) values (≤0.0157) are dominated by POM and type 2 samples, containing high proportions of PIM, have high Qbbe(547) levels (>0.0157). Qbbe(547) in offshore waters usually shows lower values than those in coastal waters (Fig. 5), i.e., the coastal areas were mainly dominated by mineral particles, whereas the offshore waters have relatively high proportions of organic particles. Moreover, the Qbbe(547) in the BS generally exhibited higher values than those in the YS.
Fig. 5. Average monthly Qbbe(547) derived from MODIS data in the BS and YS from 2003 to 2018. Contour lines indicate the threshold (0.0157) of Qbbe(547); if Qbbe(547) > 0.0157, the particles were dominated by PIM, otherwise dominated by POM.
We observed that the spatial distribution pattern of Qbbe(547) varied over time. In spring (March, April, and May), areas with high Qbbe(547) values (dominated by PIM) rapidly decreased, and correspondingly, areas with low Qbbe(547) values (dominated by POM) significantly increased. In summer (June, July, and August), low levels of Qbbe(547) became more widely distributed throughout the whole study region, and the proportion of areas with low Qbbe(547) reached maximum. Specifically, most offshore waters had low Qbbe(547), with only a few coastal areas having relatively high Qbbe(547) levels. This indicates that in summer, organic particles gradually dominate offshore waters, and inorganic particles are still the dominant particles in some coastal regions. However, during autumn (September, October, and November), the Qbbe(547) of a large area gradually shifted from low to high levels, implying that the area with high proportions of organic particles diminished, and the area with PIM-dominated particles became dominant. Nevertheless, the areas with low Qbbe(547) were still considerable offshore of the NYS and SYS, indicating that a certain proportion of water with organic-dominated particles still occupies these regions. In winter (December, January, and February), the Qbbe(547) values in most areas gradually increased to the highest level. These results demonstrate that PIM-dominated particles became dominant in most areas, except for a small portion of the offshore region that retained POM-dominated particles.
3.3 Temporal variations of backscattering efficiency and particle types
We quantitatively analyzed the temporal variations of distribution areas of POM-dominated (type 1) and PIM-dominated (type 2) particles in the whole region, BS, NYS, and SYS (Fig. 6). In the whole study area, the area fractions of PIM-dominated type 2 particles varied from 0.30 in July to 0.95 in February (average 0.59). Correspondingly, the area fractions of the POM-dominated type 1 particles changed from 0.70 to 0.05. The BS generally showed stable particle types, i.e., PIM-dominated type 2 particles most time. The highest mean value of the area fraction of type 2 particles was 0.94 (standard deviation of 0.10), and the maximum area fraction of POM-dominated particles could only reach 0.29 in July. Compared with the BS, the NYS and SYS exhibited significant variations in particle composition. The area fraction of PIM-dominated type 2 particles decreased followed by an increase, within the ranges of 0.18–0.92 and 0.27–0.95 in the NYS and SYS, respectively. The minimum area fraction mostly occurred in July or August, and the maximum values occurred in February. The factions of the POM-dominated type 1 particles showed the opposite variation pattern.
Fig. 6. Seasonal variation of the area fractions of particle types in the whole study area (a), BS (b), NYS (c), and SYS (d). Yellow and blue bars represent the area fractions of POM-dominated type 1 and PIM-dominated type 2 particles. Red lines indicate the standard deviation of the area fraction of type 1 particles.
The monthly variation in Qbbe(547) and particle types were further analyzed in the selected eight typical sub-regions, including the central BS, Bohai Strait, Yellow River Estuary, central NYS, central SYS, Subei Shoal, Yangtze Estuary, and ECS shelf (Fig. 1). In general, although Qbbe(547), and thereby particle types of these regions might be different, most of these regions showed similar monthly variation patterns (Fig. 7). The Qbbe(547) levels were generally low in summer, but high in winter. Significant differences were also observed in the Qbbe(547) values and particle types. In the central BS, Yellow River Estuary, Subei Shoal, and Yangtze Estuary, which are significantly influenced by re-suspension and transportation of mineral-rich particles introduced by river discharge or tidal forcing, Qbbe(547) showed much higher values (>0.0157) throughout the year, suggesting that the suspended particles in these areas are often PIM-dominated types. Similarly, in the Bohai Strait, the Qbbe(547) values were mostly >0.0157, indicating that PIM was the dominant particle; in August and September, Qbbe(547) dropped to slightly <0.0157, which implied that the proportions of organic particles were also considerable. However, the temporal variations in Qbbe(547) values in the central NYS, central SYS, and ECS shelf were quite different from those in the above sub-regions. The Qbbe(547) levels in the central NYS were <0.0157 (from April to December), indicating that this region mainly included POM-dominated particles. The Qbbe(547) values in the central SYS and ECS shelf were low (< 0.0157) from May to October and high (≥ 0.0157) from November to April. These results suggest that the particles in these two regions generally shifted from the POM-dominated type in summer to the PIM-dominated type in winter.
Fig. 7. Temporal variation of Qbbe(547) from January to December in eight sub-regions. Red straight lines represent the Qbbe(547) threshold for particle classification. Particles above the red line were PIM-dominated type, and those below it were POM-dominated type.
3.4 Long-term trend of particle backscattering efficiency
We further examined the long-term variation trend of the monthly Qbbe(547) anomaly and calculated the significance (p) values (Fig. 8). The Qbbe(547) in the BS showed the most apparent variation trends. Specifically, a decreasing trend was detected in the coastal area of the BS, whereas an increasing trend was observed in its offshore area. Unlike the BS, most regions of the NYS were covered by a slightly increasing trend. In the SYS area, the variation trend was heterogeneous, showing a relatively high positive trend in most coastal regions and a negative trend, mainly in offshore waters.
Furthermore, the long-term variation trends of Qbbe(547) in the eight selected sub-regions were investigated (Fig. 9). The Qbbe(547) anomaly showed clear increasing trends in the central BS and central NYS, with values of 7.63 × 10−5 (p < 0.05) and 4.69 × 10−5 (p < 0.05), respectively, implying that the proportion of PIM particles increased in the two sub-regions, and those of POM particles decreased in the past decade. In contrast, a slightly decreasing trend of the Qbbe(547) anomaly was observed in the Bohai Strait (−2.30 × 10−5, p < 0.05) and Yellow River Estuary (−8.74 × 10−6, p < 0.05), indicating a possible relative increase in the proportion of POM particles and decrease in the proportion of PIM particles in these sub-regions. The Qbbe(547) anomaly trends in the remaining four sub-regions were negligible.
Fig. 9. Linear trends of the Qbbe(547) anomaly in the eight sub-regions of the BS and YS. Red lines represent a linear trend.
4. Discussion
4.1 Rationality of the Qbbe(λ)-based model
As a crucial optical trait of suspended particles, the Qbbe(λ) is influenced by particle composition, density, and size in theory. Our earlier studies demonstrated that Qbbe(λ) has great potential for deriving information on particle composition [31,32]. Similar findings have been reported in other studies [12,13,43,44]. For instance, Neukermans et al. [44] revealed that the variability of Qbbe(λ) at 650 nm covers approximately one order of magnitude in the optically complex coastal and offshore waters around Europe and French Guyana. They stated that such variability is mainly driven by particle composition (correlation coefficient of −0.69 ± 0.12) rather than size and density. Based on in situ measurements in the Irish Sea, Celtic Sea, and English Channel, Bowers et al. [13] reported that Qbbe(λ) at 665 nm is strongly dependent on the fraction of mineral to total suspended particles, and a non-linear relationship between them was established (R2 = 0.62). These findings provide a solid theoretical base for the Qbbe(λ)-based model for particle classification.
More importantly, we reported that Qbbe(λ) can be accurately derived from satellite observations (Fig. 4). This gives the main advantage of the Qbbe(λ)-based model in terms of satellite applicability compared with the Bbp(λ)-based models, which can only be applied to in situ data in theory. When applying the Qbbe(λ)-based model to MODIS long-term data, we observed reasonable spatiotemporal variation patterns of Qbbe(547) (Fig. 5). These variation patterns generally agreed with those obtained from in situ measurements, as reported in our previous study [32], which also showed high values in the coastal region and low levels in offshore waters, and a decreasing and then increasing pattern from spring to late autumn. These findings prove that the Qbbe(λ)-based model has good satellite applicability, and provides a feasible way to study the spatiotemporal variations of Qbbe(λ) and particle composition from long-term ocean color remote sensing data.
4.2 Influencing factors of variations of Qbbe(λ) and particle types
The spatiotemporal variations of Qbbe(547) and particle types in the BS and YS were probably attributed to the dynamics of hydrographic environments, which can significantly modulate the processes of sediment re-suspension and transportation [45]. During summer, the waters in the BS and YS, especially in offshore regions, are often strongly stratified due to surface warming and low wind stress, which results in weak vertical mixing of the water column [35]. These conditions may hinder re-suspension of the mineral-rich sediment causing dominance of the POM particles and low Qbbe(547) values in surface waters. In contrast, during winter, strong northerly monsoon significantly mixes the water [35], which tends to resuspend large amounts of mineral-rich sediment. These resuspended mineral-rich sediments may further cause the dominance of PIM particles with high Qbbe(547) levels in most surface waters in the BS and YS.
In the coastal regions, apart from water mixing, sediment transport related to river discharge, tidal mixing, and mean current forcing can also cause variation in particle composition [7,46]. Freshwater discharges, especially in summer, from the Yellow River and Yangtze River, transport vast quantities of mineral-rich sediments to the BS and YS [46]. These sediment loads may explain the dominance of PIM particles in the coastal regions for most of the year (especially in the Yellow River and Yangtze River estuaries) (Figs. 5, 6, and 7). Meanwhile, tidal forcing may also be an essential factor responsible for changes in particle composition in coastal regions [7]. For instance, the PIM-dominated particles and high Qbbe (547) levels in the Subei Shoal are likely attributable to sediment re-suspension and transportation caused by tidal forcing, which is often strong in this region. Moreover, mean current forcing that can carry suspended sediment along the coastal areas of the BS and YS [47], may also be a possible factor associated with changes in particle composition.
The long-term variation in Qbbe(547) showed increasing trends in some regions, e.g., in the central BS and central NYS, and slightly decreasing trends, e.g., in the Bohai Strait and Yellow River Estuary (Figs. 8 and 9), which suggest changes in the particle composition with increasing proportions of PIM and POM particles, respectively. To further confirm these variation trends, we investigated the variability in TSM and Chla which is regarded as a proxy for phytoplankton, based on long-term MODIS data. Clear increasing trends in TSM were observed in the central BS and central NYS, while Chla has remained stable over the past decades (Figs. 10 and 11). This phenomenon implies that the increase in TSM is probably due to the increase in minerals rather than phytoplankton, which may explain the increasing trend of Qbbe(547) in these two regions. Similarly, a decreasing trend in TSM and stable Chla levels were observed in the Yellow River Estuary, suggesting a possible decrease in minerals, which has further resulted in a decreasing trend in Qbbe(547) in the area. However, slight decreases in both TSM and Chla were observed over time in the Bohai Strait. The slight decreasing trend in Qbbe(547) in the Bohai Strait may be related to the relative proportion of the reductions in TSM and Chla, whereas further investigation is required on this issue.
Fig. 10. Linear trends of TSM anomaly in the eight sub-regions of the BS and YS. Red lines represent the linear trend.
Fig. 11. Linear trends of Chla anomaly in the eight sub-regions of the BS and YS. Red lines represent the linear trend.
When extending the above analysis to the entire study region, we observed that the waters of most areas with increasing trend of Qbbe(547) generally showed increase in TSM and decrease in Chla, and vice versa (Fig. 12). The consistencies between the variation trends of Qbbe(547) and those of TSM and Chla confirm the accuracy of the long-term variability in Qbbe(547) and the particle types obtained in this study. At this stage, we must acknowledge that this was a preliminary investigation. Further investigations specifically focusing on the driving mechanisms of Qbbe(547) and particle types in the BS and YS are required based on an available dataset of both the marine environment and human activities, which may have great potential to be related to the variations of particle composition [45].
4.3 Implications and suggestions for future work
Knowledge of particle composition is essential for studying marine carbon cycles, material exchange, sediment flux, and environmental dynamics [14,22,48–50]. For instance, sinking of particulate organic carbon (POC) in the deep ocean, known as a biological pump, is thought to be significantly modulated by suspended minerals because mineral particles can increase the density of POC and protect it from microbial degradation [49,50]. Therefore, particle composition is considered to play an important role in studying marine POC flux, and thereby carbon cycles. Furthermore, Schartau et al. [22] stated that the composition of suspended particles can provide crucial information for estimating the mass exchange rates of carbon, nitrogen, phosphorus, etc., between shallow coastal waters and adjacent shelf areas. In this study, we first verified the satellite applicability of the Qbbe(λ)-based model and obtained reasonable Qbbe(λ) satellite products, from which different particle types were identified in the BS and YS. These satellite-derived products enrich and improve the detection of suspended particle properties, and may offer valuable proxies for marine biogeochemical and environmental studies. Thus, we recommend that future ocean monitoring plans include these satellite-derived products for a better understanding of the variations of particle assemblages based on long-term ocean color remote sensing data.
Although few models can derive the composition of suspended particles from satellite data, we must admit that satellite-derived particle composition information using the Qbbe(λ)-based method is qualitative, i.e., it provides the types of particles dominated by either POM or PIM rather than quantitatively estimating the proportions of different particle types. Thus, further studies are required to improve the Qbbe(λ)-based method, potentially by coupling other optical traits of the particles, such as absorption properties [51,52]. Meanwhile, although the Qbbe(λ)-based model and its performance for satellite data have been specifically studied for the waters of the BS and YS, its applicability to other coastal regions should be examined. For this, we suggest to first conduct an in-depth exploration of the relationship between Qbbe(λ) and particle composition based on field measurements, which is the optical base of this method, and then assess its performance for satellite data.
5. Conclusions
This study examined the applicability of the Qbbe(λ)-based method for particle-type classification using MODIS satellite data. Evaluations based on in situ data indicated good performance of satellite-derived Qbbe(λ) with R2, RMSE, bias, and MAPE values of 0.76, 0.007, 16.5%, and 31.0%, respectively. When applying the Qbbe(λ)-based method to long-term MODIS data, we could observe obvious spatiotemporal variations in Qbbe(λ), from which distinct particle types in the BS and YS were identified. Coastal waters showed a stable content of mineral-dominated particles for most of the year, with high Qbbe (λ) values. However, the particles in offshore waters were dominated by POM, showing low Qbbe(λ) levels in summer, and shifted to PIM-dominated type with high Qbbe(λ) values in winter. Long-term increasing and decreasing trends in Qbbe(λ) were observed, suggesting an increase in the relative proportions of PIM and POM, respectively. This was confirmed by our analysis of long-term variations in TSM and Chla. Changes in Qbbe(λ), and particle types, especially in offshore regions, were probably attributed to sediment re-suspension driven by water mixing. Moreover, sediment transportation related to river discharge and tidal forcing may also be important influencing factors in coastal waters.
Overall, this study reports the first results of the satellite application of the Qbbe(λ)-based model for identifying particle types. The satellite-derived products and obtained variation patterns of Qbbe(λ) and particle types may provide new insights and helpful knowledge for better studying marine biogeochemical processes in the BS and YS. Further investigations are required to focus on the applicability of the Qbbe(λ)-based model to other regions and its improvement to derive quantitative information (i.e., the relative proportions) of different particle types in future studies.
Funding
National Natural Science Foundation of China (41876203, 42106176, 42176179, 42176181); Natural Science Foundation of Jiangsu Province (BK20210667, BK20211289); Open Fund of State Key Laboratory of Satellite Ocean Environment Dynamics (QNHX2243); Open Fund of State Key Laboratory of Remote Sensing Science (OFSLRSS202103); National Key Research and Development Program of China (2021YFC2803301, 2022YFC3106003); Open Fund of Key Laboratory of Coastal Zone Development and Protection (2021CZEPK02); NSFC Open Research Cruise funded by Shiptime Sharing Project of NSFC. (NORC2018-01).
Acknowledgments
We acknowledge Ocean Biology Processing Group (OBPG) at NASA's Goddard Space Flight Center for providing MODIS-Aqua satellite data.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. R. F. Anderson and C. T. Hayes, “Characterizing marine particles and their impact on biogeochemical cycles in the GEOTRACES program,” Prog. Oceanogr. 133, 1–5 (2015). [CrossRef]
2. M. Bellacicco, M. Cornec, E. Organelli, R. Brewin, G. Neukermans, G. Volpe, M. Barbieux, A. Poteau, C. Schmechtig, and F. d’Ortenzio, “Global variability of optical backscattering by non-algal particles from a biogeochemical-Argo data set,” Geophys. Res. Lett. 46(16), 9767–9776 (2019). [CrossRef]
3. H. Xi and Y. Zhang, “Total suspended matter observation in the Pearl River estuary from in situ and MERIS data,” Environ. Monit. Assess. 177(1-4), 563–574 (2011). [CrossRef]
4. B. Paudel, P. A. Montagna, and L. Adams, “The relationship between suspended solids and nutrients with variable hydrologic flow regimes,” Reg. Stud. Mar. Sci. 29, 100657 (2019). [CrossRef]
5. E. Organelli, G. Dall’Olmo, R. J. Brewin, G. A. Tarran, E. Boss, and A. Bricaud, “The open-ocean missing backscattering is in the structural complexity of particles,” Nat. Commun. 9(1), 5439 (2018). [CrossRef]
6. C. Mobley, The Oceanic Optics Book, 1st ed. (International Ocean Colour Coordinating Group, 2022).
7. X. He, Y. Bai, D. Pan, N. Huang, X. Dong, J. Chen, C.-T. A. Chen, and Q. Cui, “Using geostationary satellite ocean color data to map the diurnal dynamics of suspended particulate matter in coastal waters,” Remote Sens. Environ. 133, 225–239 (2013). [CrossRef]
8. D. Doxaran, N. Lamquin, Y.-J. Park, C. Mazeran, J.-H. Ryu, M. Wang, and A. Poteau, “Retrieval of the seawater reflectance for suspended solids monitoring in the East China Sea using MODIS, MERIS and GOCI satellite data,” Remote Sens. Environ. 146, 36–48 (2014). [CrossRef]
9. M. Jafar-Sidik, F. Gohin, D. Bowers, J. Howarth, and T. Hull, “The relationship between Suspended Particulate Matter and Turbidity at a mooring station in a coastal environment: consequences for satellite-derived products,” Oceanologia 59(3), 365–378 (2017). [CrossRef]
10. Y. Xiang and P. J. Lam, “Size-fractionated compositions of marine suspended particles in the Western Arctic Ocean: Lateral and vertical sources,” J. Geophys. Res.: Oceans 125(8), e2020JC016144 (2020). [CrossRef]
11. J. Wei, M. Wang, L. Jiang, X. Yu, K. Mikelsons, and F. Shen, “Global estimation of suspended particulate matter from satellite ocean color imagery,” J. Geophys. Res.: Oceans 126(8), e2021JC017303 (2021). [CrossRef]
12. D. Stramski, E. Boss, D. Bogucki, and K. J. Voss, “The role of seawater constituents in light backscattering in the ocean,” Prog. Oceanogr. 61(1), 27–56 (2004). [CrossRef]
13. D. Bowers, P. Hill, and K. Braithwaite, “The effect of particulate organic content on the remote sensing of marine suspended sediments,” Remote Sens. Environ. 144, 172–178 (2014). [CrossRef]
14. M. J. Behrenfeld, E. Boss, D. A. Siegel, and D. M. Shea, “Carbon-based ocean productivity and phytoplankton physiology from space,” Global Biogeochem. Cy. 19(1), 1 (2005). [CrossRef]
15. B. D. Walker, S. R. Beaupré, T. P. Guilderson, M. D. McCarthy, and E. R. Druffel, “Pacific carbon cycling constrained by organic matter size, age and composition relationships,” Nat. Geosci. 9(12), 888–891 (2016). [CrossRef]
16. T. L. Cucci, S. E. Shumway, W. S. Brown, and C. R. Newell, “Using phytoplankton and flow cytometry to analyze grazing by marine organisms,” Cytometry 10(5), 659–669 (1989). [CrossRef]
17. E. S. Poloczanska, M. T. Burrows, C. J. Brown, J. García Molinos, B. S. Halpern, O. Hoegh-Guldberg, C. V. Kappel, P. J. Moore, A. J. Richardson, and D. S. Schoeman, “Responses of marine organisms to climate change across oceans,” Front. Mar. Sci. 62 (2016).
18. A. Ody, D. Doxaran, Q. Vanhellemont, B. Nechad, S. Novoa, G. Many, F. Bourrin, R. Verney, I. Pairaud, and B. Gentili, “Potential of high spatial and temporal ocean color satellite data to study the dynamics of suspended particles in a micro-tidal river plume,” Remote Sens. 8(3), 245 (2016). [CrossRef]
19. X. Zhang, C. G. Fichot, C. Baracco, R. Guo, S. Neugebauer, Z. Bengtsson, N. Ganju, and S. Fagherazzi, “Determining the drivers of suspended sediment dynamics in tidal marsh-influenced estuaries using high-resolution ocean color remote sensing,” Remote Sens. Environ. 240, 111682 (2020). [CrossRef]
20. Q. Wang, Y. Li, and Y. Wang, “Optimizing the weight loss-on-ignition methodology to quantify organic and carbonate carbon of sediments from diverse sources,” Environ. Monit. Assess. 174(1-4), 241–257 (2011). [CrossRef]
21. P. J. Lam, J.-M. Lee, M. I. Heller, S. Mehic, Y. Xiang, and N. R. Bates, “Size-fractionated distributions of suspended particle concentration and major phase composition from the US GEOTRACES Eastern Pacific Zonal Transect (GP16),” Mar. Chem. 201, 90–107 (2018). [CrossRef]
22. M. Schartau, R. Riethmüller, G. Flöser, J. van Beusekom, H. Krasemann, R. Hofmeister, and K. Wirtz, “On the separation between inorganic and organic fractions of suspended matter in a marine coastal environment,” Prog. Oceanogr. 171, 231–250 (2019). [CrossRef]
23. M. S. Twardowski, E. Boss, J. B. Macdonald, W. S. Pegau, A. H. Barnard, and J. R. V. Zaneveld, “A model for estimating bulk refractive index from the optical backscattering ratio and the implications for understanding particle composition in case I and case II waters,” J. Geophys. Res.: Oceans 106(C7), 14129–14142 (2001). [CrossRef]
24. H. Loisel, X. Mériaux, J.-F. Berthon, and A. Poteau, “Investigation of the optical backscattering to scattering ratio of marine particles in relation to their biogeochemical composition in the eastern English Channel and southern North Sea,” Limnol. Oceanogr. 52(2), 739–752 (2007). [CrossRef]
25. X. Zhang, R. H. Stavn, A. U. Falster, D. Gray, and R. W. Gould Jr, “New insight into particulate mineral and organic matter in coastal ocean waters through optical inversion,” Estuarine, Coastal Shelf Sci. 149, 1–12 (2014). [CrossRef]
26. E. Boss, L. Guidi, M. J. Richardson, L. Stemmann, W. Gardner, J. K. Bishop, R. F. Anderson, and R. M. Sherrell, “Optical techniques for remote and in-situ characterization of particles pertinent to GEOTRACES,” Prog. Oceanogr. 133, 43–54 (2015). [CrossRef]
27. D. Sun, X. Su, S. Wang, Z. Qiu, Z. Ling, Z. Mao, and Y. He, “Variability of particulate backscattering ratio and its relations to particle intrinsic features in the Bohai Sea, Yellow Sea, and East China Sea,” Opt. Express 27(3), 3074–3090 (2019). [CrossRef]
28. W. Zhou, W. Cao, J. Zhao, G. Wang, W. Zheng, L. Deng, and C. Li, “Variability of scattering and backscattering of marine particles in relation to particle concentration, size distribution, and composition off the eastern hainan coast in the south China sea,” Cont. Shelf Res. 232, 104615 (2022). [CrossRef]
29. Z. Lee, K. L. Carder, and R. A. Arnone, “Deriving inherent optical properties from water color: a multiband quasi-analytical algorithm for optically deep waters,” Appl. Opt. 41(27), 5755–5772 (2002). [CrossRef]
30. T. J. Smyth, G. F. Moore, T. Hirata, and J. Aiken, “Semianalytical model for the derivation of ocean color inherent optical properties: description, implementation, and performance assessment,” Appl. Opt. 45(31), 8116–8131 (2006). [CrossRef]
31. S. Wang, Z. Qiu, D. Sun, X. Shen, and H. Zhang, “Light beam attenuation and backscattering properties of particles in the Bohai Sea and Yellow Sea with relation to biogeochemical properties,” J. Geophys. Res.: Oceans 121(6), 3955–3969 (2016). [CrossRef]
32. S. Wang, S. Chen, Z. Qiu, D. Sun, H. Zhang, W. Perrie, and T. Zhang, “Variability in the backscattering efficiency of particles in the Bohai and Yellow Seas and related effects on optical properties,” Opt. Express 24(26), 29360–29379 (2016). [CrossRef]
33. S. Wang, Y. Huan, Z. Qiu, D. Sun, H. Zhang, L. Zheng, and C. Xiao, “Remote sensing of particle cross-sectional area in the Bohai Sea and Yellow Sea: algorithm development and application implications,” Remote Sens. 8(10), 841 (2016). [CrossRef]
34. D. Sun, Y. Huan, Z. Qiu, C. Hu, S. Wang, and Y. He, “Remote-Sensing Estimation of Phytoplankton Size Classes From GOCI Satellite Measurements in Bohai Sea and Yellow Sea,” J. Geophys. Res.: Oceans 122(10), 8309–8325 (2017). [CrossRef]
35. C.-T. A. Chen, “Chemical and physical fronts in the Bohai, Yellow and East China seas,” J. Mar. Syst. 78(3), 394–410 (2009). [CrossRef]
36. J. Sündermann and S. Feng, “Analysis and modelling of the Bohai sea ecosystem—a joint German–Chinese study,” J. Mar. Syst. 44(3-4), 127–140 (2004). [CrossRef]
37. M. Zhang, J. Tang, Q. Song, and Q. Dong, “Backscattering ratio variation and its implications for studying particle composition: A case study in Yellow and East China seas,” J. Geophys. Res. 115(C12), 1 (2010). [CrossRef]
38. R. A. Maffione and D. R. Dana, “Instruments and methods for measuring the backward-scattering coefficient of ocean waters,” Appl. Opt. 36(24), 6057–6067 (1997). [CrossRef]
39. Y. Agrawal and H. Pottsmith, “Instruments for particle size and settling velocity observations in sediment transport,” Mar. Geol. 168(1-4), 89–114 (2000). [CrossRef]
40. Z. Lee, K. L. Carder, and R. A. Arnone, “Update of the Quasi-Analytical Algorithm (QAA_v6),” https://www.ioccg.org/groups/Software_OCA/QAA_v6_2014209.pdf (2014).
41. E. Siswanto, J. Tang, H. Yamaguchi, Y.-H. Ahn, J. Ishizaka, S. Yoo, S.-W. Kim, Y. Kiyomoto, K. Yamada, and C. Chiang, “Empirical ocean-color algorithms to retrieve chlorophyll-a, total suspended matter, and colored dissolved organic matter absorption coefficient in the Yellow and East China Seas,” J. Oceanogr. 67(5), 627–650 (2011). [CrossRef]
42. G. Zibordi, F. Mélin, J.-F. Berthon, and E. Canuti, “Assessment of MERIS ocean color data products for European seas,” Ocean Sci. 9(3), 521–533 (2013). [CrossRef]
43. E. Flory, P. Hill, T. Milligan, and J. Grant, “The relationship between floc area and backscatter during a spring phytoplankton bloom,” Deep Sea Res., Part I 51(2), 213–223 (2004). [CrossRef]
44. G. Neukermans, H. Loisel, X. Mériaux, R. Astoreca, and D. McKee, “In situ variability of mass-specific beam attenuation and backscattering of marine particles with respect to particle size, density, and composition,” Limnol. Oceanogr. 57(1), 124–144 (2012). [CrossRef]
45. S. Qiao, X. Shi, G. Wang, L. Zhou, B. Hu, L. Hu, G. Yang, Y. Liu, Z. Yao, and S. Liu, “Sediment accumulation and budget in the Bohai Sea, Yellow Sea and East China Sea,” Mar. Geol. 390, 270–281 (2017). [CrossRef]
46. X. Liu, L. Qiao, Y. Zhong, X. Wan, W. Xue, and P. Liu, “Pathways of suspended sediments transported from the Yellow River mouth to the Bohai Sea and Yellow Sea,” Estuarine, Coastal Shelf Sci. 236, 106639 (2020). [CrossRef]
47. H. Ichikawa and R. C. Beardsley, “The current system in the Yellow and East China Seas,” J. Oceanogr. 58(1), 77–92 (2002). [CrossRef]
48. T. S. Arnarson and R. G. Keil, “Organic–mineral interactions in marine sediments studied using density fractionation and X-ray photoelectron spectroscopy,” Org. Geochem. 32(12), 1401–1415 (2001). [CrossRef]
49. J. D. Wilson, S. Barker, and A. Ridgwell, “Assessment of the spatial variability in particulate organic matter and mineral sinking fluxes in the ocean interior: Implications for the ballast hypothesis,” Global Biogeochem. Cy. 26(4), 2012GB004398 (2012). [CrossRef]
50. W. Deng, L. Monks, and S. Neuer, “Effects of clay minerals on the aggregation and subsequent settling of marine Synechococcus,” Limnol. Oceanogr. 60(3), 805–816 (2015). [CrossRef]
51. G. Neukermans, R. A. Reynolds, and D. Stramski, “Optical classification and characterization of marine particle assemblages within the western Arctic Ocean,” Limnol. Oceanogr. 61(4), 1472–1494 (2016). [CrossRef]
52. D. Liu, H. Duan, S. Yu, M. Shen, and K. Xue, “Human-induced eutrophication dominates the bio-optical compositions of suspended particles in shallow lakes: Implications for remote sensing,” Sci. Total Environ. 667, 112–123 (2019). [CrossRef]
